Impact of MOVES2014 on Emission Inventories from On-road Mobile Sources Alexis Zubrow, Darrell Sonntag, Harvey Michaels, David Brzezinski, Alison Eyth

MOVES2014 EPA’s mobile emission model Replaces the MOVES2010 series Used for State Implementation Plan and Transportation Conformity Analysis Replaces the MOVES2010 series Estimates emissions and energy use from all on-road sources at the national, county, and project scales Contains the impacts of recent regulations: Light-duty and Heavy-duty Greenhouse Gas and Fuel Economy Standards Tier 3 Vehicle Emissions and Fuel Standards Program (emission standards and sulfur content in gasoline) Contains new emissions and activity data for on-road vehicles Adds nonroad emission capabilities through the incorporation of NONROAD2008

New Emission Test Programs & Data Fuel Effects EPAct study on Gasoline fuel effects (funded by DOE, CRC and EPA) Effects of E85 on emissions EPA In-Use Sulfur Test Program Updated renewable fuel usage and future fuel supply Evaporative Emissions Ethanol, RVP, leak magnitude study High Evap Field Study – leak frequency Running loss vapor leak study Multiday Diurnal Testing PM Emissions Speciation of Kansas City light-duty gasoline study New PM Speciation profiles for diesel and CNG from literature Temperature Effects EPA Cold Temperature Study Heavy Duty In-Use Compliance Program emissions data Literature review of CNG Transit bus emissions Population and Activity R.L. Polk Vehicle Population for 2011 Updated Vehicle Sales Projections Updated FHWA VMT estimates Updated Flex-fuel vehicle penetrations National average speed distribution using GPS data Updated truck weights based on Weigh-in- Motion data Note for Alexis: These are just some of the major updates. We had many more updates, which make it difficult to attribute changes to which update.

Air Quality Modeling using MOVES2014 VOC and PM2.5 emission rates by model year, fuel type, regulatory class, emission process Speciation incorporated into MOVES2014 Formerly conducted within SMOKE through COMBO files MOVES2014 estimates CB05 mechanism species 18 PM2.5 species needed for CMAQv5 Aerosol Module Version 6 and CAMx5.4 Allows differentiation in VOC and PM speciation profiles by: Technology/Regulatory class (e.g. light-duty/heavy-duty) Model year (e.g. pre-2007/2007+ diesel) Fuel Type (gasoline, E85, diesel, CNG) Emission process (e.g. running, start, extended idle, evaporative) Plan to update supported mechanisms to include CB6 and other mechanisms SPECIATE PM2.5 and VOC Speciation Profiles SMOKE Prepares emissions in gridded, hourly form needed for Chemical Transport Model MOVES produces 55 toxics, EC, sulfate makes 57 unique species i.e. account for mix of pre- and post-2007 diesel No model year distinction, fuel distinction, emission process distinction in SCC Custom weighted speciation profiles need to be applied for each calendar year Chemical Transport Model

3-city comparisons State-supplied National Emission Inventory (NEI) inputs for 3 urban counties Unique fleet mix, fuels, temperatures, age distributions, inspection & maintenance programs, average speed distributions, road types, etc. Projections to 2030 using national default growth assumptions Consistent inputs were developed for both MOVES2010b and MOVES2014 Note: certain fleet characteristics are different between the models, e.g. fuel splits

Highway-Source NOx Relative change with MOVES2014 2011 Reduced truck weights Updated gasoline fuel effects 2030 Large changes due to emission and fuel regulations (Tier 3 and HD GHG reductions)   NOx 2011 -21% to -5% 2030 -57% to -53%

Highway-Source VOC Relative change with MOVES2014 2011 Updates to evaporative and fuel effects for gasoline vehicles 2030 Reductions in 2017+ vehicles (Tier 3 Low Sulfur rule) Increase in temperature sensitivity for newer gasoline vehicles (MY 2010+)   VOC 2011 -21% to -5% 2030 -33% to -13%

Highway-Source PM2.5 Relative change with MOVES2014   PM 2011 -19% to -6% 2030 -58% to -42% 2010 PM running emissions less sensitive to temperature (lowers emissions esp. in cold weather 2030 Tier 3 reductions

3 State Comparison Comparison of 2011 NEIv1 vs 2011 NEIv2 (draft) for 3 states GA, IA, MD Cumulative impact from change in model from MOVES2010b to MOVES2014 AND changes in inputs Using MOVES in emissions rate mode (SMOKE-MOVES)

Updated Inputs Inputs revised for 2011 NEIv2 (draft): Age distribution New default county specific age distribution for light duty vehicles (LDV), from CRC A-88 Regional long-haul splits Changes split between short vs long-haul for the single unit (52, 53) and combination (61, 62) trucks, from CRC A-88 Impacts hotelling Regional fuels Representative counties Previously 163 representative counties, now 284 within CONUS domain Revised age bins and added ramp fraction Includes state comments

Updated Inputs (2) Input changes (cont) : Hotelling Revised CDBs Includes extended idle and auxiliary power units (APU) Distributed based on urban + rural restricted combination long- haul VMT (62) New type of run: rate-per-hour (RPH) that takes hotelling hours directly Limited number of states supplied hotelling hours Revised CDBs Some states supplied updated CDBs for 2011 NEIv2. Temporalization Incorporated monitoring data VTRIS with state submitted MOVES CDB data Spatial surrogates Meteorology

3 State PM10 increasing across all states: change in brake and tire wear EF

3 State: Broad Vehicle NOx VMT Shift in Iowa VMT from passenger -> trucks (primarily LDT) based on improved population inputs Increase in Iowa diesel “All trucks and buses” emissions primarily from changes in hotelling (EXT and APU)

Observations With equivalent inputs, MOVES2014 estimates lower emissions in urban areas Larger reductions typically observed for future years Due to projected impact of EPA regulations Revised inputs can create greater changes in the emissions than changes to MOVES alone Shifts in vehicle distribution can have subtle impacts Setting up “equivalent” inputs is difficult for SMOKE-MOVES New source classification codes

Next Steps Compare results across more scenarios Run AQ sensitivities Compare SMOKE-MOVES for a change in MOVES alone with “consistent” inputs Evaluate future year impacts

Acknowledgments EPA: Laurel Driver, Rich Cook, Chris Dresser, David Choi, Ted Maciag (SEE) ERG: Allison DenBlyker, John Koupal, Scott Fincher CSC: Chris Allen, James Beidler UNC: BH Baek, Catherine Seppanen MOVES MJO Workgroup

Extra 3-City Slides

3-city comparisons Consistent Input Data between MOVES2010b and MOVES2014 State-supplied data for the National Emission Inventory Base year (2011) Vehicle Miles Traveled and vehicle population Vehicle Age distribution (2011) Average speed distribution Road type distribution (e.g. urban vs. rural) Temporal VMT distribution (vehicle travel by hour of day, day of week, month of year) Inspection/Maintenance Program Common Projections to 2030 National VMT growth rates National vehicle population growth rates Modeled future changes to age distributions (based on 2011 state supplied age distributions) MOVES default Data Meteorology data (same between runs) Fuel Formulation and Fuel Supply data from MOVES2014 (did not specify E85 fuel usage in 2010b) Differences between runs Emission rates, fuel effects, temperature effects, “Internal default vehicle classifications in MOVES” Fuel Split (e.g. diesel, gasoline, CNG E85) by vehicle model year and vehicle type Mileage accumulation rates by vehicle classes Vehicle weights Default extended idle activity

Differences in 3 State Inputs Extended idle MOVES2010b used vehicle population (VPOP) of heavy- duty diesel vehicles (SCC7 2230074) to calculate extended idle emissions MOVES2014 uses hotelling hours directly to calculate extended idle emissions SCCs MOVES2010b uses MOBILE6 based SCCs MOVES2014 uses codes more inline with MOVES IDs. No 1-1 mapping between 2 types of SCCs. To compare: Map fuels to all non-diesel or diesel Map source types to: motorcycles, passenger cars, all trucks and buses. Within “all trucks and buses”, MOVES2014 has a different composition of the fleet

Differences in 3 State Inputs (cont.) Activity data No clean mapping between SCCs -> difficult to match VMT and VPOP exactly between two runs Equivalent at the aggregate level (e.g. All trucks and buses) Distribution within fuels and within source type may be different in the two runs